Friday, March 30, 2012

From Measuring Poverty to poverty of measurement? Source:

Methodological slips in the poverty estimation exercise in India has reached a juncture where one is reminded of the point made by Karl Marx that history repeats itself, first as tragedy and then as farce (The Eighteenth Brumaire of Louis Napoleon). The tragedy occurred in the late nineties, precisely in 1997, when the government decided to divide the population into Below Poverty Line (BPL) and Above Poverty Line (APL) in order to target subsidized food, based on poverty lines which had been wrongly (under)estimated over the past two decades. Subsequently, it was discovered that the poverty lines, based on declared nutritional norms (and hence was accepted as an appropriate tool for targeting food subsidies), have since long deviated from the prescribed norms in the original methodology. This methodological slip proved tragic for millions who were in need for subsidized food from the Public Distribution System but were excluded on the basis of the under-estimated poverty lines, multiplying their chances to be gripped in a state of chronic hunger and wasted life.

 Once the original methodological gaffe was unmasked, an expert group was constituted by the government to review the poverty estimation methodology, popularly known as the Tendulkar Committee (TC henceforth). The new methodology proposed by the TC in 2009 has now been accepted by the government and the 2009-10 poverty estimates recently published (33.2% for rural and 20.9% for urban India) are based on this changed methodology. Unfortunately, far from the correcting the original methodological error, the TC in its bid to rescue the poverty estimates from the awkward clutches of nutritional requirements of the population, ended up rationalizing and providing an apology for the underestimation of poverty in the country. In the process, a more fundamental methodological goof-up is committed, this time in the very approach to poverty measurement. The TC methodology has now officially made poverty estimation normless, thereby rendering the new poverty estimates as a farce. Needless to say, the other problem of non-comparability of poverty lines over time and space has persisted with this detachment of any ‘norm’ to poverty estimates.
At the first instance, the proposition that the TC has build up an apology for underestimation of poverty may seem odd to many given that their estimates for rural India for 2004-5 was much higher at 41.8% than the corresponding 28.3% estimated by the old methodology. However if we study the actual problems in the estimation process of the old methodology and the inappropriate way in which the TC has tried to address that, this proposition will be clearer to us.
Shortcomings of the old methodology
The major caveat of the planning commission poverty lines was gross underestimation of poverty with the degree of underestimation amplifying with time. The original norm of 2400 Kcal per day per capita for rural areas and 2100 Kcal per day per capita for urban areas was applied by the 1979 Task force of the planning commission for the year 1973-74 to estimate poverty. The corresponding Monthly Percapita Consumption Expenditure (MPCE) levels for the 1973-74 NSS round at which these norms can be achieved was designated as the poverty line and the proportion of the population below that line gave the corresponding poverty estimates. The poverty lines and estimates for the subsequent rounds of NSS consumption expenditure surveys (normally every 5 years) were determined by updating the 1973-74 poverty lines by price indices (CPI-AL and CPI-IW), separately for rural and urban areas. In actuality, the calorie norm originally applied in 1973-74 was less than the recommended cut-off. It was actually 2200 Kcal for rural areas and 2000 Kcal for urban areas, as shown by subsequent enquiries (Patnaik, 2007, 2010), though this was never made official.
What has happened over the years is that the successive poverty lines have progressively deviated from the original calorie norm on which they were based. So much so that by 2004-5, the calorie intake possible at the official poverty line was only 1820 Kcal per capita per day; 380 Kcal below the original 2200 Kcal. Similar deviations occurred for urban areas also and by 2004-5, the poverty line allowed a calorie intake of only 1835 Kcal; 165 Kcal less than the original 2000 Kcal (ibid). The fact that the poverty lines of the different years (and also across states) embodied varying normative energy intakes implies that the corresponding poverty estimates were temporally (and spatially) non-comparable. In other words, as the yardstick for determining whether a person is poor or not kept changing for the different rounds, the declining trend in poverty estimates that can be observed in successive rounds is a specious trend. In actuality, it is brought about by lowering the norm for every successive exercise of poverty estimation.
The poverty estimates have been rendered spurious over time primarily due to the mechanical fashion in which the CPI-AL and the CPI-IW was used for rural and urban areas to update poverty lines. This indirect method of generating successive poverty lines by updating the original line solely by a general price index has major weaknesses, which depleted the scientific and technical credibility of the poverty estimates over the last three decades.
The consumption basket that is used for poverty estimation remains unchanged over the successive rounds of poverty estimation over three decades. This has no rationale as there are continuous structural changes in the economy which are not captured by a general price index. The latter can only help in accounting for the factor of inflation within a period of time. A mere updating of poverty lines accounting only for inflation does not account for the changes in the intrinsic character of the consumption basket itself over time.
The change in the character of the basket occurs precisely due to numerous structural changes in the economy which general price indices cannot capture. For example, say an agricultural labourer receives a part of her wages in kind, in terms of rice valued at farm gate prices. The remaining rice that she requires for her household is procured at a higher retail market price. Thus, the average price at which she consumes rice is a weighted average of the farm gate and retail prices. When the same person starts receiving her wages entirely in cash at some later point of time, it means that she has to buy her entire requirement of rice from the market at the retail price, which would be definitely higher than the earlier weighted average price that she was paying. The same amount of money that she earlier spent on rice consumption would allow her to procure a lesser physical quantity of rice with lesser embodiment of calories. Given that she still wants to consume the same amount of rice which allows her to intake the same amount of energy that she has been doing earlier, she has to ceteris paribus increase the share for rice in her total consumption budget. One should note that all this can happen even when the price of rice in the economy remains invariant.
Similar such structural changes that continually occur within an economy imply that the ‘true’ changes in the cost of living are not captured by prices alone. Thus, it is amply clear that the more appropriate way to capture poverty is by directly using certain norms regarding basic essentials like food, clothing, shelter, etc. and by sticking to the norms over specific time periods. The direct method of assessing levels of consumption relative to the ‘norm’ in different years implicitly takes into account the structural changes in the economy, which the official indirect method had missed out. The calorie norm is one such benchmark and similar norms for other commodities can also be used to account for them explicitly, though they had been implicitly accounted for in the original exercise in 1973-4.
Tendulkar Committee: More of the same fallacies?
The new methodology for poverty estimation suggested by the TC somewhat recognizes the shortcomings of the earlier methodology discussed above but commits certain fundamental errors at the same time. The committee has now suggested a new Poverty Line Basket (PLB) instead of any fixed norm. The urban national poverty line for 2004-05 estimated by the earlier methodology has been recommended as the new PLB. The earlier practice of differential norm for rural and urban areas has been replaced by this universal PLB for the whole economy. This All-India urban PLB is then projected for the various states to obtain the state-level urban poverty line using national-state price differentials. The state-specific rural-urban price differentials are then used to estimate the rural poverty lines for the various states. Finally, the rural All-India poverty line is estimated as a weighted average of state rural poverty lines (using population shares as weights).
The revised rural poverty line for 2004-05 was MPCE Rs. 446.7 below which 41.8 per cent of the rural population lay. This has now been accepted by the government. For all India in 2004-5, the revised poverty estimate is 37.2 percent. The TC methodology when used on the 2009-10 NSS Consumption data (Mixed Recall Period) gives us the 2009-10 poverty lines for rural and urban India; an MPCE of Rs. 672.8 (rural) and Rs. 859.6 (urban). The corresponding poverty estimates are 33.2% in rural India and 20.9% in urban India. This indeed reveals a massive decline in poverty, 8% in rural and 4.8% (from 25.7% in 2004-05) in urban India, regarding which the ruling establishment in this country has gone euphoric!
Why and how are these poverty estimates and the corresponding declines that they exhibit purely bogus? The answer lies in studying a set of fundamental errors committed and mythical assumptions adopted by the TC in developing the new methodology. The major weakness of the TC recommendations lies in the domain of fixing the PLB for poverty line determination. The committee has stated that it is purposefully moving away from the ‘calorie anchor’ of poverty estimation citing problems with the nutrition data published by the NSS. Instead, they have chosen the urban poverty line basket for 2004-05 as the base PLB for both urban and rural areas. There are at least three problems that can be identified with this choice and the rationale behind it.
First, in its attempt to move away from the calorie norm, the new methodology actually moved away from fixing any norm for poverty measurement in the country. Instead, the TC has fixed a particular consumption basket as the PLB. This essentially embodies a norm-less measurement of poverty in the country. Any consumption basket is only a ‘tool’ that represents a certain norm or benchmark for measuring poverty. Replacing a norm by the tool itself for measuring poverty implies a continuation and perhaps intensification of the earlier arbitrariness that has plagued poverty estimation. The answer to the problem of using an unchanging consumption basket for estimating the poverty line (old methodology) that increasingly got detached from a declared benchmark is not to change the consumption basket arbitrarily, but to ensure that the consumption basket changes so as to represent some fixed norm of poverty measurement at different points of time. Without the latter, poverty estimates are not comparable over time.
Not only does this simple logic escape the reason of the TC, but by abolishing the very use of norms for poverty measurement, the TC has circumvented the shortcomings of the earlier methodology without addressing them. In case the committee believed that the calorie norm is inappropriate for measuring poverty (more on this later), it would have been useful if they had suggested some new and improved norm without altogether ignoring the fundamental dimension of hunger in poverty.
Secondly, the argument that the nutrition data of the NSS has problems and it is better to measure poverty indirectly using the MPCE figures is a myth. The NSS collects data on the monthly consumption of different food and non-food items in its Consumption Expenditure Survey (CES). It then uses the unit prices for these items to reach the monthly per capita expenditure (MPCE) figures which are published in the CES. On the other hand, the same physical quantities of food items when multiplied by their unit calorie values provide the nutritional intake of the population. There is not much scope of error in the unit calorie values of different food items, given they are scientifically determined and not amenable to any significant change over time. If at all there is some error, it has to be with the consumption survey itself, in which case the PLB based on MPCE data which is chosen as the new benchmark will also be affected by some such errors.
Thirdly and most importantly, the choice of the urban poverty line for 2004-05 as the new PLB for determining the poverty line is grossly erroneous. The committee betrays a certain amount of naiveté when it mentions ‘The estimated urban share of the poor population […] in 2004-05 […] is generally accepted as being less controversial than its rural counterpart […] that has been heavily criticized as being too low’ (GOI, 2009). When both the rural and urban poverty lines have been estimated by the same erroneous methodology, it is generically not possible that one can be an underestimation even as the other capturing poverty accurately. Patnaik (2010) shows that for the urban areas, the poverty line accounted for only 1835 Kcal, 265 Kcal (165 Kcal from 2000 Kcal) below the prescribed norm and the proportion of people below 2100 Kcal in 2004-05 in urban areas was 64.2 percent, 38.5 percent above the official estimate of 25.7 percent.
Fixing this urban poverty line basket of 2004-05, which has been derived after systematically underestimating poverty levels for three decades, as the new starting point for measuring poverty by the TC effectively means that the latter has forwarded an apology for the earlier errors as well as made an official endorsements of the wrongs that has been done to the poor of this country. Indeed, the TC methodology is not for poverty estimation but a method for ‘systematically underestimating poverty’.
One can meaningfully stop this argument at this point and debunk the new poverty estimates arrived at by the TC methodology. However, the hollowness of the claims and rationalization made by the TC regarding its methodology deserve not to go without comment.
An important claim of the TC is that by shifting from the use of general price indices like the CPI-AL/CPI-IW to ‘cost of living’ indices (CoL), it has prevented the systemic and progressive underestimating of poverty that the earlier methodology was essentially leading to. The CoL indices are constructed using the household level unit values for different commodities/commodity groups (as many as 23 groups). This reveals the misplaced obsession with an indirect and cumbersome method of estimating poverty using some kind of price or other, when many direct indicators are available (not only nutritional intakes but also indicators of health, housing, etc) that can be used for estimating the poor population.
The claim that CoL indices actually account for the structural changes in the consumption patterns of the population is also not foolproof. Take the following example: A household shifts from consumption of rice/wheat to small millets over time due to stress in their earnings. The CoL index constructed for cereals will be unable to account for this change. Small millets which have a much lower price and unit value (but also much lower calories embodied in them) will keep the CoL index for cereals at a lower level even as the total calories embodied in this changed cereal basket will be much lower than the rice/wheat basket. Any change in quality of goods consumed cannot be captured by the CoL based on the unit values. Thus, with the existing caveat of prices not capturing structural changes accurately enough persisting with the CoL indices, one can raise a question regarding the adoption of this much more complicated methodology for estimating poverty.
The other important claim of improvement in the methodology is the inclusion of education and health expenditures explicitly within the PLB. A closer scrutiny of the method of updating poverty lines over time reveals that these crucial items such as education and health are clubbed together with other items like entertainment, etc. into a ‘miscellaneous’ group and this latter group is updated by the general price indices as was done in the old methodology (GOI, 2009, page 25-26). In the absence of any prescribed acceptable norms for education or health status at the poverty line, it is more probable that within a short span of time, these expenditures at the poverty line would be such that cannot account for any decent quantum of education or health (similar to what happened with nutritional attainments at the old poverty lines).
Reading the Tendulkar Committee report, one perceives a sense of ambiguity amongst the authors after deriving the poverty lines and estimates. This is reflected in the multiple rationalizations that they resort to in order to prove their estimated poverty line as realistic. In this convoluted exercise of rationalization, what are the parameters that they fall back upon? Surprisingly, the report first takes refuge in that so-called ‘irrelevant’ (and already rejected by the TC itself) parameter of ‘calorie intake’ at the poverty line.
The TC’s own findings reveal that the urban PLB for 2004-05 can provide for 1776 Kcal per capita per day. The FAO Minimum Dietary Energy Requirement (MDER) for India for 2004-6 (1770 Kcal) is quoted by them to justify this very low calorie intake at the new urban poverty line. In such an argument, it is ignored that the MDER specified by the FAO is by definition the minimum dietary intake required by the population ‘for maintaining a healthy life and carrying out a light physical activity’ (emphasis added by author, Glossary, FAO). If one checks the FAO document ‘Country Profile: Food Security Indicators’ for India (available at the following link ‘ accessed on 24 March 2012), which gives these MDER figures, one cannot miss from the next line of information that the FAO also conceptualizes an ‘Average Dietary Energy Requirement’ (ADER) higher than the MDER (summarized for various years in Table 1). The ADER specifies the ‘amount of dietary energy per person that is considered adequate to meet the energy needs for heavy activity and good health’ (ibid).
1: FAO MDER and ADER (in Kcal) for India: various years
By ignoring the ADER, which is more appropriate for a society where large segments of the population is engaged in hard physical labour in agriculture, mining, construction or the exploitative informal economy, the TC has betrayed its objective of manipulating living standards for the purpose of proving their estimates as correct. The poverty lines that have been reached at can thus be designated as some measure which is irrelevant for the poor, working population of this country, perhaps more relevant for the urban sedentary office worker. Utter disdain for the labour of the working poor which has created the riches of this society is wholesomely reflected in this rationalization by the committee.
Several studies have noted that a higher dietary intake is required for maintaining a healthy life taking into account men and women performing hard physical labour in the economy. The Challenge of Hunger 2007 published by the IFPRI rightly notes, ‘FAO’s measure of the proportion of the population with calorie deficiency is, however, based on the minimum and not the average dietary energy requirements, and therefore it cannot capture the proportion of people whose calorie supply lies between these two thresholds’ (IFPRI 2007, Part I, Endnote 13, Page 20). The report also notes that the average norm is particularly crucial for South Asia, and specifically India, where a significantly large proportion of the population lies between these two thresholds (ibid, Page 10). Weisman and Smith (2007) classified the deprivation below the two norms (minimum and average) as ‘Food Energy Deficiency’ and ‘Severe Food Energy Deficiency’. Patnaik (2007) has termed the proportion of the population below 1800 Kcal as the ‘depth of poverty’ even as she measured the incidence of poverty using higher norms.
The other rationalization tool used by the TC is the set of ex-post external validation checks whether the actual expenditures on food, education and health at the poverty line can allow decent standards of consumption of these items. This is where the TC finally remembers that poverty estimation must satisfy some ‘norm’. The adequacy tests carried out for 2004-05 show that at the poverty line, the food expenditures are satisfied for 17 and 19 states (out of the 20 major states) respectively for rural and urban areas. What needs to be noted is that the parameter used for arriving what food expenditure is adequate is the Body Mass Index (BMI). The relevance of BMI (ratio of weight to square of height) for measuring adequacy is questionable in a country where there is a widespread incidence of child under-nourishment and malnutrition. The latter not only causes wasting (low body-weight) but also stunting (low height) rendering the BMI as largely misleading (also pointed out by Patnaik, 2010).
The TC does not give the information on how many states satisfied the adequacy tests for education and health but Himanshu (2010) provides that. The latter reveals that among the 20 major states, only 12 and 10 states in rural and urban respectively had education and health expenditures at the poverty line that satisfied 90% of the normative expenditure. Fewer states actually reach the norm. This also raises serious doubts over how properly education and health expenditures were actually accounted for by the TC methodology, one of their important claims of improvement.
An apology is instantaneously forwarded that these shortfalls at the poverty lines are on account of hospitalization and not due to education or non-institutional treatment (ibid, GoI, 2009, page 9). Of course, we are supposed to remember here that in the age of neo-liberalism, the poor working population is not only ‘not entitled’ to adequate food-energy but are also not expected to tread into a hospital when they fall sick from over-working and under-nutrition!  
The 2009-10 poverty lines: the critique validated
Much of our anticipated criticism of the Tendulkar Committee methodology is substantiated by the new poverty estimates for 2009-10 that the Planning Commission has recently published. Using the NSS data on nutrition and consumption from the Nutritional Intakes in India, 2009-10 (Report no. 540) and Level and Pattern of Consumption Expenditure, 2009-10 (Report No. 538) and applying the direct methodology (Patnaik, 2007), we can estimate the nutritional intakes that are allowed by the new poverty lines. The rural poverty line for 2009-10 (MPCE Rs. 672.8) allows a calorie intake of 1880 Kcal per capita per day; 119 Kcal less than 1999 Kcal attainable at the 2004-05 rural poverty line. There is a similar decline in the calorie intake possible at the urban poverty line for 2009-10 (MPCE Rs. 859.6) to 1730 Kcal per capita per day; 46 Kcal less than the already low 1776 Kcal attainable at the 2004-05 PLB.
This decline not only implies that the poverty lines are rendered non-comparable over time due to this change in the associated material well-being parameters but also validates our criticism that complicated CoL indices are still incapable of capturing structural changes in the economy accurately. The problem is amplified now for the Planning Commission and the ruling establishment in this country as their earlier practice of invoking the FAO MDER figures is also not an option any more. Table 1 clearly shows that both the MDER and ADER figures published by FAO has increased since 2004-06 (and also before this period), while the poverty line dietary intakes have declined in 2009-10 compared to 2004-05.
The other rationalization usually put forward is that higher calorie intakes are usually possible at the official poverty lines but it is only the ‘voluntary’ diet diversification of the poor which leads them to purchase lower calories at the poverty line. This apology also falls flat upon preliminary speculation of the NSS nutrition data for 2004-05 and 2009-10 (NSS Report No. 513 and 540). The per capita per day protein consumption for the 2004-05 rural poverty line (Rs. 446.7) MPCE class was 54.1 grams; the same figure for the 2009-10 rural poverty line class (a different class now) is only 51.4 grams. For the same class (41-50% of the population) which is now non-poor in 2009-10, the protein intake figure is still lower at 53.3 grams in 2009-10. Clearly people near the rural poverty line are consuming lesser proteins (or definitely not increasing) also along with lesser dietary energy intakes. In urban areas, the protein intake at the poverty line class in 2004-05 was 55.3 grams per capita per day; the figure for the poverty line class (the same class in this case, 20-30% of the population) in 2009-10 is substantially lower at 47.9 grams!
Generally the average protein and fat intakes for the whole population has also declined between 2004-05 and 2009-10, the only exception being the fat intake in urban areas, where the figures has increased by a mere 0.4 grams. One can comprehend therefore what are the dimensions of diet diversification occurring in India that many frequently cite as an apology for falling food energy intakes! No amount of rationalization of the poverty lines can stand in the face of the reality that rising hunger-incidence in India in not only about calorie-deprivation but of a serious decline in all nutritional indicators. With cereals still accounting for 64.8% and 56.4% of the protein intakes in this country in 2009-10 (NSS Report 540), it is only natural that the declining food-grains availability under the neo-liberal policy regime has seriously jeopardized the food-intakes of large segments of the population and handicapped them in their hourly battle against hunger.
This author’s estimates show that 74.9 percent of the rural population lies below a per capita per day calorie intake of 2200 Kcal (less than the FAO prescribed ADER) in 2009-10. In urban areas, 60.7 percent of the population has a calorie intake of less than per capita per day 2000 Kcal in 2009-10. In other words, the planning commission poverty lines leave out roughly 42 per cent and 40 per cent of the population, who are actually food-deprived, in rural and urban India respectively, from the scope of their definition of the poor. Hunger has little correlation with poverty is what we learn from the Planning Commission experts!
Any honest and transparent effort of poverty estimation should have urged the Planning Commission to at least publish the external validation checks (though even these checks are inappropriate according to this author) for 2009-10 that the original exercise carried out for 2004-05. Given the arbitrary manner in which these adequacy tests were formulated, it is most probable that their linkages with the new poverty lines have gone haywire and the authorities are in no position to publish them. But without publishing these checks, the planning commission has no justification for publishing the new poverty estimates if the body believes in minimum transparency.
No matter what the ruling establishment tells us, there is simply no case for accepting the government’s poverty estimates and their spurious methodology for such estimates. Not only should these claptrap poverty figures be not linked to any public social welfare programmes, more so with food distribution, but the erroneous poverty estimation methodology should be shelved altogether. The official poverty estimates are only of use to those who have the unilateral objective of restricting public expenditures by reducing the BPL estimates and for those who press for the continuation and speeding up of the neo-liberal economic reforms. From the point of view of the people of this country, there is no relevance for these misleading poverty estimates. Their urgent need is for universal social welfare programmes that cater to their basic needs.
The unwarranted and erroneous division of large sections of the vulnerable population, mostly poor and laboring people, into ill-founded categories of BPL and APL is a cruel joke that the ruling classes are playing on them. However, the Indian rich elite of this country must not ignore the message that a poster conveyed at a recent Occupy Wall Street demonstration-‘One day the poor will have nothing left to eat but the rich’. The reasons and pre-conditions of our own Occupy Movement are here. 
Government of India (1979) Report of the Task Force on Projections of Minimum Needs and Effective Consumption Demand, Planning Commission, New Delhi, January.
_________ (2009) Report of the Expert Group to Review the Methodology for Estimation of Poverty, Planning Commission, New Delhi, November.
_________ (2012) Nutritional Intakes in India, 2009-10, NSS Report no. 540, Ministry of Statistics and Programme Implementation
_________ (2012) Level and Pattern of Consumption Expenditure, 2009-10, NSS Report no. 538, Ministry of Statistics and Programme Implementation
_________ (2007) Nutritional Intakes in India, 2004-5, NSS Report no. 513, Ministry of Statistics and Programme Implementation
Food and Agricultural Organization,
Himanshu (2010) ‘Towards New Poverty Lines for India’ Economic and Political Weekly, Vol. 45(1), pp: 38-48.
International Food Policy Research Institute (2007) Global Hunger Index: Facts, Determinants and Trends, The Challenge of Hunger 2007 report prepared in collaboration with Welt Hunger Hilfe and Concern, Bonn, October.
Patnaik Utsa (2007) ‘Neoliberalism and Rural Poverty in India’ Economic and Political Weekly, Vol 42(30), pp: 3132-3150.
Patnaik Utsa (2010) ‘Trends in Urban Poverty under Economic Reforms: 1993-94 to 2004-05’ Economic and Political Weekly, Vol 45(4), pp: 42-53.
Smith, Lisa C. and Doris Weismann (2007) ‘Is Food Insecurity More Severe in South Asia or Sub-Saharan Africa?’, IFPRI Discussion Paper 712, IFPRI, Washington.
Top of Form

No comments:

Post a Comment

Thank you for your comments